Home / Careers / ML Engineer (Audio & Speech)
AUDIO & AI TEAM
ML Engineer (Audio & Speech)
Train and deploy models for source separation, speech recognition, voice similarity and music understanding.
Ahmedabad · Hybrid
Full-time
3+ years
Posted 8 Nov 2026
About the role
You will train, evaluate and serve audio and speech models, from licensed datasets to GPU inference, for music-tech and media clients.
What you'll do
✓
Train and fine-tune models for separation, speech and voice embeddings
✓
Build audio datasets with clear licensing records
✓
Run evaluations with SDR, WER and listening tests
✓
Optimise models for GPU serving and cost
✓
Work with DSP engineers on audio pre-processing
What you bring
✓
3+ years in ML, including audio or speech work
✓
Strong PyTorch and Python
✓
Experience with audio features and augmentation
✓
Models deployed to production, not only notebooks
✓
Clear experiment tracking habits
Nice to have
✓
Demucs, Whisper or similar model experience
✓
ONNX, TensorRT or quantisation
✓
Published work or open-source contributions
What we offer
✓
Paid certification exams and study time
✓
Learning budget for courses and conferences
✓
Health cover for you and your family
✓
Hybrid working from Zundal, Ahmedabad
Team
Audio & AI
Location
Ahmedabad, India · Hybrid
Type
Full-time
Experience
3+ years
Salary
Competitive, shared on first call
Your hiring manager
Audio DSP Lead, Audio & AI team
WHAT HAPPENS NEXT
From application to offer in about 2 weeks
01
Apply
2 minutes
02
Intro call
30 minutes
03
Practical test
A short lab, not homework
04
Meet the team & offer
1 hour
APPLY
Apply for ML Engineer (Audio & Speech)
SIMILAR ROLES
You might also like
What does a ML Engineer (Audio & Speech) do at Zuroth?
You will train, evaluate and serve audio and speech models, from licensed datasets to GPU inference, for music-tech and media clients. Train and deploy models for source separation, speech recognition, voice similarity and music understanding.
What experience do I need for this role?
We look for 3+ years in ML, including audio or speech work, strong PyTorch and Python and experience with audio features and augmentation. If you meet most of the requirements and are keen to learn, we still want to hear from you.
Is this role hybrid or remote?
This role is hybrid from our office in Zundal, Ahmedabad, with flexible days agreed with your manager. Most client work is for companies in the United States and Canada, so some days include a short overlap with US Eastern Time.
What does the hiring process look like?
There are four steps: a 2-minute application, a 30-minute intro call, a short practical exercise relevant to the role, and a one-hour meeting with the team before an offer. Most candidates hear back within 3 business days and finish in about 2 weeks.
What growth and learning support do you offer?
Every role has a learning budget for courses and conferences, paid certification exams with study time, and a growth path reviewed twice a year. Senior roles mentor others and help shape how our practices work.